Table 3. Analysis of combined spectrum and inoculum size effect on sensitivity of rapid antigen detection test (RADT).

	Univariate analysisa							Multivariate analysisa
Variable	All inocula	p	Light inoculum	p	Heavy inoculum	p	p b	Light inoculumc	p	Heavy inoculumd	p
	(n  = 285)		(n  = 42)		(n  = 243)			(n  = 42)		(n  = 243)
Age (years)
3–8	ref	0.01	ref	<0.01	ref	0.49	0.03	ref	<0.001	ref	0.36
9–14	−22.4 (−39.7; −5.1)		−37.4 (−64.1; −10.7)		−4.6 (−17.4; 8.3)			−40.9 (−57.9; −24.0)		−5.7 (−17.9; 6.4)
Lack of cough
Yes	ref	0.03	ref	0.09	ref	0.14	0.19	ref	<0.001	ref	0.08
No	−10.0 (−19.1; −9.0)		−25.0 (−53.6; 3.6)		−5.2 (−12.1; 1.7)			−23.6 (−23.7; −23.5)		−6.0 (−12.5; 0.6)
Anterior cervical adenopathy
No	ref	0.07	ref	0.01	ref	0.88	0.01	ref	0.01	ref	0.37
Yes	7.3 (−0.6; 15.2)		36.1 (7.7; 64.5)		−0.4 (−6.1; 5.2)			40.5 (16.7; 64.3)		2.0 (−2.3; 6.2)
Tonsillar swelling or exudate
No	ref	0.40	ref	0.12	ref	0.21	0.07	n/ae		ref	0.29
Yes	4.3 (−5.6; 14.3)		−25.0 (−56.2; 6.2)		5.4 (−3.1; 13.9)					4.0 (−3.4; 11.4)
Fever
No	ref	0.61	ref	0.27	ref	0.77	0.26	n/ae		ref	0.41
Yes	−2.3 (−10.8; 6.3)		−19.1 (−53.0; 14.8)		1.0 (−5.6; 7.6)					2.3 (−3.3; 8.0)

Data RADT sensitivity difference (95% confidence interval) unless indicated.

Abbreviations: ref, reference; n/a, not applicable.

^aBinomial modelling.

^bInteraction test between clinical variables and inoculum size (Wald test).

^cModel adjusted on age, lack of cough, anterior cervical adenopathy.

^dModel adjusted on age, lack of cough, anterior cervical adenopathy, tonsillar swelling or exudate and fever.

^eNot included in the model because of non-convergence.